NSCT域内基于引导滤波与改进PCNN的CT/MRI医学图像融合方法  

CT/MRI Fusion Method Based on Guided Filtering and Improved PCNN in NSCT Domain

作　　者：于淼 宁春玉[1] 石乐民 吕冰垚 YU Miao;NING Chun-yu;SHI Le-min;LV Bing-yao(School of Life Science and Technology,Changchun University of Science and Technology,Changchun 130022)

机构地区：[1]长春理工大学生命科学技术学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2020年第5期137-142,共6页Journal of Changchun University of Science and Technology(Natural Science Edition)

基　　金：吉林省科技发展计划项目(20180623008TC)。

摘　　要：针对CT和MRI图像融合边缘模糊、有伪影等问题,提出了一种改进引导滤波与自适应脉冲耦合神经网络(Pulse Coupled Neural Network,PCNN)相结合的图像融合方法。首先,对源图像通过非下采样Contourlet变换(Non-sub⁃sampled Contourlet Transform,NSCT)得到低通子带和带通子带。然后对低通子带采用自适应的PCNN进行融合。其中,用改进的平均梯度作为连接强度;用改进的拉普拉斯能量和作为外部激励;点火映射图的判决遵循取大原则。对于带通子带采用改进的引导滤波进行融合。最后,通过NSCT逆变换得到融合结果图。多组CT和MRI图像融合实验结果表明,此算法能更多地保留源图像的信息,边缘保持能力更强。融合图像对比度高,视觉效果更佳,在CT和MRI医学图像融合方面效果显著。Aiming at the problems of blurred edges and artifacts in CT and MRI image fusion,an improved guided filtering and PCNN image fusion was proposed.Firstly,t he source image was transformed by non-subsampled Contourlet transform to get the lowpass sub-band and the bandpass sub-band.Secondly,the lowpass sub-band was fused by adaptive PCNN.The improved average gradient was used as the connection strength;and the improved sum-modified Laplacianwas usedas an external excitation.The judgment of fire mapping image followed the maximum principle.The improved guided filter was adopted to fuse the bandpass sub-band.Finally,the fusion image was obtained by NSCT inverse transformation.Fusion experiments were conducted on multiple groups of CT and MRI images.It is proved that this algorithm can retain more information of source image and has stronger edge preserving ability.The fusion image has high contrast and better visual effect.In the aspect of CT and MRI medical image fusion,the effect is remarkable.

关 键 词：图像处理 医学图像融合 自适应PCNN 引导滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]